Spaces:
Sleeping
Sleeping
Upload 6 files
Browse files
app.py
CHANGED
|
@@ -51,11 +51,16 @@ def upload_pdf(files):
|
|
| 51 |
|
| 52 |
|
| 53 |
|
|
|
|
|
|
|
|
|
|
| 54 |
def generate_qa(token):
|
| 55 |
try:
|
| 56 |
if not token:
|
| 57 |
return "⚠️ Please provide a token."
|
| 58 |
|
|
|
|
|
|
|
| 59 |
# Load chunk_data using token
|
| 60 |
with sqlite3.connect("my_database.db") as conn:
|
| 61 |
cursor = conn.cursor()
|
|
@@ -63,41 +68,59 @@ def generate_qa(token):
|
|
| 63 |
row = cursor.fetchone()
|
| 64 |
|
| 65 |
if not row:
|
|
|
|
| 66 |
return "❌ No data found for this token."
|
| 67 |
|
| 68 |
chunks = json.loads(row[0])
|
|
|
|
|
|
|
|
|
|
|
|
|
| 69 |
qa_pairs = []
|
| 70 |
|
| 71 |
-
for chunk in chunks:
|
|
|
|
| 72 |
questions = qgen.generate(chunk)
|
|
|
|
|
|
|
| 73 |
if not questions:
|
|
|
|
| 74 |
continue
|
| 75 |
|
| 76 |
for question in questions[:2]: # Max 2 Qs per chunk
|
| 77 |
prompt = f"Context: {chunk}\n\nQuestion: {question}\n\nAnswer:"
|
|
|
|
|
|
|
| 78 |
try:
|
| 79 |
result = qa_model(prompt, max_length=256, do_sample=False)
|
|
|
|
|
|
|
| 80 |
if isinstance(result, list) and "generated_text" in result[0]:
|
| 81 |
answer = result[0]["generated_text"].strip()
|
| 82 |
elif isinstance(result, dict) and "answer" in result:
|
| 83 |
answer = result["answer"].strip()
|
| 84 |
else:
|
| 85 |
answer = "N/A"
|
|
|
|
|
|
|
| 86 |
qa_pairs.append(f"Q: {question}\nA: {answer}")
|
|
|
|
| 87 |
except Exception as e:
|
| 88 |
-
print(f"QA model failed: {e}")
|
| 89 |
continue
|
| 90 |
|
| 91 |
if not qa_pairs:
|
|
|
|
| 92 |
return "⚠️ No Q&A pairs generated."
|
| 93 |
|
|
|
|
| 94 |
return "\n\n".join(qa_pairs)
|
| 95 |
|
| 96 |
except Exception as e:
|
|
|
|
| 97 |
return f"❌ Error: {str(e)}"
|
| 98 |
|
| 99 |
|
| 100 |
-
|
| 101 |
# ✅ Ask question using token (semantic similarity)
|
| 102 |
def ask_question(token, question):
|
| 103 |
try:
|
|
|
|
| 51 |
|
| 52 |
|
| 53 |
|
| 54 |
+
# Load QG and QA once
|
| 55 |
+
qgen = QGenerator()
|
| 56 |
+
qa_model = pipeline("text2text-generation", model="google/flan-t5-base")
|
| 57 |
def generate_qa(token):
|
| 58 |
try:
|
| 59 |
if not token:
|
| 60 |
return "⚠️ Please provide a token."
|
| 61 |
|
| 62 |
+
print("📥 Received Token:", token)
|
| 63 |
+
|
| 64 |
# Load chunk_data using token
|
| 65 |
with sqlite3.connect("my_database.db") as conn:
|
| 66 |
cursor = conn.cursor()
|
|
|
|
| 68 |
row = cursor.fetchone()
|
| 69 |
|
| 70 |
if not row:
|
| 71 |
+
print("❌ No data found for token in DB.")
|
| 72 |
return "❌ No data found for this token."
|
| 73 |
|
| 74 |
chunks = json.loads(row[0])
|
| 75 |
+
if not chunks:
|
| 76 |
+
print("⚠️ Chunk data is empty.")
|
| 77 |
+
return "⚠️ No content available in database for this PDF."
|
| 78 |
+
|
| 79 |
qa_pairs = []
|
| 80 |
|
| 81 |
+
for i, chunk in enumerate(chunks):
|
| 82 |
+
print(f"\n🔹 Processing chunk {i+1}/{len(chunks)}")
|
| 83 |
questions = qgen.generate(chunk)
|
| 84 |
+
print(f"🧠 Questions generated: {questions}")
|
| 85 |
+
|
| 86 |
if not questions:
|
| 87 |
+
print("⚠️ No questions generated for this chunk.")
|
| 88 |
continue
|
| 89 |
|
| 90 |
for question in questions[:2]: # Max 2 Qs per chunk
|
| 91 |
prompt = f"Context: {chunk}\n\nQuestion: {question}\n\nAnswer:"
|
| 92 |
+
print(f"➡️ Prompt:\n{prompt}")
|
| 93 |
+
|
| 94 |
try:
|
| 95 |
result = qa_model(prompt, max_length=256, do_sample=False)
|
| 96 |
+
print(f"⬅️ Raw model output: {result}")
|
| 97 |
+
|
| 98 |
if isinstance(result, list) and "generated_text" in result[0]:
|
| 99 |
answer = result[0]["generated_text"].strip()
|
| 100 |
elif isinstance(result, dict) and "answer" in result:
|
| 101 |
answer = result["answer"].strip()
|
| 102 |
else:
|
| 103 |
answer = "N/A"
|
| 104 |
+
|
| 105 |
+
print(f"✅ Final Answer: {answer}")
|
| 106 |
qa_pairs.append(f"Q: {question}\nA: {answer}")
|
| 107 |
+
|
| 108 |
except Exception as e:
|
| 109 |
+
print(f"❌ QA model failed: {e}")
|
| 110 |
continue
|
| 111 |
|
| 112 |
if not qa_pairs:
|
| 113 |
+
print("⚠️ No Q&A pairs generated.")
|
| 114 |
return "⚠️ No Q&A pairs generated."
|
| 115 |
|
| 116 |
+
print("✅ Final Q&A generated successfully.")
|
| 117 |
return "\n\n".join(qa_pairs)
|
| 118 |
|
| 119 |
except Exception as e:
|
| 120 |
+
print(f"🔥 Exception in generate_qa(): {e}")
|
| 121 |
return f"❌ Error: {str(e)}"
|
| 122 |
|
| 123 |
|
|
|
|
| 124 |
# ✅ Ask question using token (semantic similarity)
|
| 125 |
def ask_question(token, question):
|
| 126 |
try:
|